Physical Modelling from a circuit schematic

[DavenH]

Yes.. and thanks, but what's immensely frustrating is that the included song isn't the correct version... if I'm not mistaken, there's a missing sample, and all the Sytruses in that song load the default preset! (I was testing the beta at the time the installer was put together)

Dude, that song was so brilliant I almost cried...And that was the bad version?! Is there a place where I could get the correct version?

I hope Reflex will update it in the next beta. I am honored to have it included, make no mistake! It's just that it's unplayable with those Sytruses having the default patch, which is quite cpu intensive, and the missing sample means it takes forever to load. You can get the proper version here (note that you'll need FL v6.3 to run this). Thanks for the praise!

Learjeff - thanks for your comments. Certainly natural playing style is a big task, but that's not something I need to worry about just yet

[james0tucson]

How does one go about developing a software model of an electronic circuit?

The circuit simulators don't come anywhere close to the realtime performance needed for audio. Even the really expensive circuit modelling systems don't touch it.

[AdmiralQuality]

Exactly what learjeff said!

Daven, can you post an MP3? I'd like to check it out.

[DavenH]

I suppose now that most points have been made it doesn't hurt to go a little off topic

Tesseract

Keep in mind that the samples used are FL native, as with the effects.

[AdmiralQuality]

Shameless self promotion is what keeps KVR alive my friend.

Listening now... not my thing, but good anyway! Keep it up.

[DavenH]

Shameless self promotion is what keeps KVR alive my friend.

Listening now... not my thing, but good anyway! Keep it up.

Cheers!

[VladimirDimitrievich]

I suppose now that most points have been made it doesn't hurt to go a little off topic

Tesseract

Keep in mind that the samples used are FL native, as with the effects.

Brilliant!

[Zyxoas]

A few points. Actual scientific physical modelling exist I say! Remember the Yamaha VL1? Wait, remember the FRUITY PLUCKED!? They both use a form of PM named digital waveguide synthesis where they model the actual waves using delay lines (I believe there's a Wikipedia article on it). It was "invented" by Julius Orion Smith, the same guy who invented the DX7 FM algorithm (but the simple plucked algorithm was invented much earlier by 2 guys named Karplus and Strong). This is also how Slayer works (the guitar part - not the amp). It's for emulating blown and plucked instruments though, not amps. Download a demo of Tassman4 by AAS. Some systems are impossible to emulate perfectly naively digitally in real-time for various reasons (such as the "delay-free loop" in the Moog filter). There's no point working on a guitarifier - that's not how real amps work. And you should really stop being so stubborn! You're too smart for that, or is the money you're getting from IL tainted with Gol-serum?

[AdmiralQuality]

Hahahaaa. Yeah, I dunno why but those Fruity folks are a surly bunch.

BTW, I thought it was John Chowning at Stanford who invented FM synthesis? http://en.wikipedia.org/wiki/FM_synthesis



And yes, there's some KINDA-SORTA physical modelling out there. But like I said, until someone models the actual physics, in a piecewise simulation (and there's lots of software that's doing this in various R&D applications, but usually not in real time) PM is just a way to describe algorithms that are designed to make one particular kind of sound. For example one of the many VSTi electric piano algo's vs. a generalized synthesis method like FM that's suited to making all kinds of different sounds.)

I want to be able to model a 3 inch tall tuba, made out of plastic, played on the surface of Mars. Or a musical instrument made from the guy wires on a radio tower (Star Wars blaster style!). Or a mile high pipe organ. Etc...

I'm hoping to get into some PM experiments myself after the Poly-Ana project winds down. Some things I'd like to take on are:

- A convincing spring reverb, approached as a piece-wise sim. So each spring would be hundreds or even thousands of elements, all interacting.
- A room reverb that actually models the pressure in the air, in 3D.
- A convincing tape and/or oil can delay effect.
- I used to want to do a Wurlitzer electric piano, but then I heard the free Mr. Ray and think that market is now destroyed, for the time being at least.
- And yes, even an pedal/amp sim!

[learjeff]

Zyxoas makes a good point. Sometimes physical modeling doesn't have to be complicated. For example, you don't have to model every molecule in a volume of gas since PV = nRT. If you're only interested in the pressure, modeling every molecule would be a waste of processing. Transmission lines are like that too: they're well understood in theory, and theory and practice tend to agree quite strongly.

I'd say it's not "kinda-sorta" modeling, but it is modeling limited to relatively simple physical phenomena. Or, more accurately, modeling physical phenomena for which relatively simple models are very accurate.

[AdmiralQuality]

I was using "molecule" above as a descriptive term, for the smallest piece that you'd simulate. No, you wouldn't ever need to actually simulate something down to molecules (or the count of the number of extra electrons stored in a capacitor, from another of my "why we don't" examples.)

So say you were doing a 3D room reverb. Maybe you'd divide the room up into thousands of little cubes, and yes, they'd have pressure, and that pressure would propagate according to the pressure of the neighbouring cube/molecules. And... it would all work out... maybe?

As is, reverbs pretend sound travels in a discreet number of rays, which is a DRASTIC oversimplification. And often they only work in a 2D space. Yes, they can sound pretty good anyway... but like I said, I just wanna try it someday. (I will probably find its prohibitive from the processing power and RAM it would require... but I'd consider doing an offline one too, just to hear it.)

S'all.

[learjeff]

convolutions don't use rays; convolutions are pretty darn good simulations. The biggest weakness there is creating the impulse and recording it. Starter pistols do pretty well to create one, so the weak link is probably the mike.

[learjeff]

BTW, if you want response up to 20kHz, the wavelength is .68 inches or so, and you'd need to model considerably smaller than your smallest wavelength. So, say half-inch on a side at the biggest. For a 20x40x12 room, that's about 22 million cubes ... good luck doing that in real time!

[AdmiralQuality]

Convolutions aren't simulations at all learjeff. This my the point folks. Just because it *emulates* something, doesn't make it a *simulation*, nor even a *physical model*.

And yes, I've thought about the number of cubes. Like I said, "I will probably find its prohibitive from the processing power and RAM it would require... but I'd consider doing an offline one too, just to hear it." So I know... it'd be unlikely to work in real time. But computers are only going to get faster. Maybe I'll just have some code in the can for a few years, waiting for people to get Hal9000s on their desks.

Just for fun, I used to code solar system simulations, and they're a REAL bitch because each object's gravity acts on every other object! Processing increases exponentially with planet population. So I have a good feel for this stuff, I know its got nasty CPU requirements.

But there's ways to make the cube primitives bigger. I think. (Keeping these ideas to myself for now.)

Anyway, just throwing this out as an example of something I'd consider an actual physical model. And it also illustrates wonderfully why that kind of modelling usually ISN'T the way to go. And that when you see the term "physical modelling", it doesn't actually mean they modelled physics.

[Zyxoas]

I might be wrong about JOS, but he's also from Stanford (CCRMA) so it's not entirely impossible that he was involved with FM. One could emulate a room using digital waveguide synthesis in a mesh-like structure. You can even do drum heads with digital waveguides! As I said before, download Tassman4. String Studio, also from AAS, is another incredible example of the power digital waveguide synthesis. Perhaps, once you've tried them out, those two examples might convince you that there's more to physical modelling than you currently believe.